#1255 perf(vk): the dead render targets go at every present, as RX_VK_DEAD_EVERY=1 meant, and a pooled image waits 8 presents (RX_VK_POOL_IDLE)
Commit
Plan (gpt-6.1-terra)
Linux market camera turns: compare render-target defaults and earlier behavior against main
In plain words
On Linux, the plan teleports a new save to Castle Town's market and turns the camera, comparing PR defaults and the earlier render-target behavior against main. It takes roughly 30 minutes per tester and measures VRAM, frame rate and frame times, with snapshots for visual regressions. Crashes, substantial VRAM or performance regressions, and picture differences fail the plan. The requested town save is unavailable; render-target block sizes and transient flicker cannot be directly compared by these probes.
Asked for
linux: Castle Town's market with the camera turning (save set "First Step into Hyrule Town", the walk north into the market, then the camera turned for a minute), once with this PR's defaults and once with `RX_VK_DEAD_EVERY=10 RX_VK_POOL_IDLE=120` (the behaviour before it). Expecting: no crash; the same picture on both, with no flicker of shadows or lit windows; with the defaults less video memory in use and smaller render-target blocks (the log's `vk: VRAM:` and `image blocks (render targets)` lines), and a frame rate no worse. 🤖 Generated with [Claude Code](https://claude.com/claude-code)
Raw plan Download plan.json
{
"pass": {
"crash": false,
"fps_mean": {
"min_delta_pct": 5
},
"screenshot": {
"max_diff": 0.05
},
"vram_peak_mb": {
"max_delta_pct": 10
},
"frametime_p99_ms": {
"max_delta_pct": 15
}
},
"probe": [
{
"kind": "crash"
},
{
"kind": "vram"
},
{
"kind": "frametime"
},
{
"kind": "draws"
},
{
"kind": "log_match",
"expected": [],
"forbidden": [
"vk: device lost"
]
}
],
"matrix": {
"pr_only": {
"RX_VK_POOL_IDLE": [
"120"
],
"RX_VK_DEAD_EVERY": [
"10"
]
}
},
"schema": 3,
"target": [
{
"name": "linux",
"tags": [
"linux"
],
"min_reports": 1
}
],
"summary": "Linux market camera turns: compare render-target defaults and earlier behavior against main",
"baseline": {
"ref": "main"
},
"scenario": [
{
"env": {},
"name": "market-defaults",
"runs": 1,
"step": [
{
"do": "teleport",
"to": "Market_Child"
},
{
"do": "wait",
"map": "Market_Child",
"event": "map",
"timeout_s": 60
},
{
"do": "shot",
"name": "market-start",
"upload": false,
"compare": "baseline"
},
{
"do": "repeat",
"step": [
{
"x": 1,
"y": 0,
"do": "stick",
"stick": "R",
"frames": 600
}
],
"count": 3
},
{
"do": "shot",
"name": "market-mid",
"upload": false,
"compare": "baseline"
},
{
"do": "repeat",
"step": [
{
"x": 1,
"y": 0,
"do": "stick",
"stick": "R",
"frames": 600
}
],
"count": 3
},
{
"do": "shot",
"name": "market-end",
"upload": false,
"compare": "baseline"
},
{
"do": "measure",
"name": "market",
"warmup_s": 0,
"duration_s": 20
}
],
"start": "new",
"matrix": false,
"settings": {}
},
{
"env": {},
"name": "market-previous",
"runs": 1,
"step": [
{
"do": "teleport",
"to": "Market_Child"
},
{
"do": "wait",
"map": "Market_Child",
"event": "map",
"timeout_s": 60
},
{
"do": "shot",
"name": "market-start",
"upload": false,
"compare": "baseline"
},
{
"do": "repeat",
"step": [
{
"x": 1,
"y": 0,
"do": "stick",
"stick": "R",
"frames": 600
}
],
"count": 3
},
{
"do": "shot",
"name": "market-mid",
"upload": false,
"compare": "baseline"
},
{
"do": "repeat",
"step": [
{
"x": 1,
"y": 0,
"do": "stick",
"stick": "R",
"frames": 600
}
],
"count": 3
},
{
"do": "shot",
"name": "market-end",
"upload": false,
"compare": "baseline"
},
{
"do": "measure",
"name": "market",
"warmup_s": 0,
"duration_s": 20
}
],
"start": "new",
"matrix": true,
"settings": {}
}
]
}
Targets
| Target | Tier | Reports | Status |
|---|---|---|---|
| linux | reviewed GummiNaomi | none yet | 0/1 pass, 0 running, 0 queued |
Reports
No reports yet. Each tester posts one when its job ends.
Jobs
| Job | Commit | Target | State | Tester | Started | Attempts | Estimate | Note |
|---|---|---|---|---|---|---|---|---|
jb_8205f7b75dc5e3d3 log | f490c2a | linux | cancelled build-base | gab9281/agent-236e29 | 1 | the PR was closed |